Introduction: The measurement of regional differences in consumer price levels is important to policy makers in business, government and academics as well as to individual citizens faced with decisions on where to live. Estimates of the magnitude of regional price differences are needed in comparisons of real income, levels of living or consumer expenditure patterns across regions. In large Federal countries such as India and the US, with considerable heterogeneity in preferences, quality of items and household characteristics between regions, the calculation of regional price differentials, hence, acquires considerable importance. There is, therefore, a significant literature, mostly based on US data, on the measurement of regional cost of living [see, for example, Moulton (1995), Kokoski, Moulton and Zeischang (1999), Koo, Phillips and Sigalla (2000)]. The literature on multilateral price index numbers is mostly concerned with the construction of PPP’s/exchange rates from item/group-wise price and quantity/expenditure/ share data available at the level of region/country. There is no reference to the use of microlevel data (for example, household level data on commodity prices/unit values available from countrywide consumer expenditure surveys) for estimation of multilateral price index numbers reflecting regional price differentials. However, given the fact that such micro-level data often contain valuable price information, it is worth while to explore if such data can be utilised to measure regional price differentials by estimating multilateral (consumer) price index numbers when the data set covers households belonging to more than one region (namely, districts within a region, states/provinces within a country or a set of countries). The purpose of the present paper is to attempt and report such an exercise. To be precise, given a set of cross-sectional household level expenditure data obtained from a nation-wide survey, we consider the subset of items/item groups for which household level price/unit value and quantity measurements are both available. We then specify a price equation [ie., a ‘quality equation’ in the terminology of Prais and Houthakker (1955)] for each of these items/item groups by relating its price/unit value to the household’s level of living (as measured by the household’s per capita total consumer expenditure (PCE)) and a set of relevant household attributes (for example, household age-sex composition) together with two sets of dummies – one set relating to the items/products and the other set relating to the regions. The proposed methodology employs a two-stage estimation procedure. In the first stage, the item/item group-wise price equations are estimated and, hence, the region-wise estimates of slope and of the intercepts of the item-specific price equations are obtained. In the second stage, the set of multilateral regional price index numbers are estimated by regressing the region-specific intercept differentials on the corresponding slope differentials of individual items/item groups using another dummy variables based regression equation. This procedure is closely related to the CPD methodology because the price equation described above shares the hedonic feature which is central to the idea of the CPD model. There is, however, a basic difference – viz., we use the household PCE and attributes as surrogates for quality of items/item groups consumed by a sample household, rather than the information of item quality (which is usually not recorded in great detail in consumer expenditure surveys). The paper is organised as follows: Section 2 specifies the price equation, explains via a reference to the CPD model the rationale of the proposed regression based procedure for estimating multilateral price index numbers from household level price/unit value data (Section 2.1) and describes the estimation method (Section 2.2). Section 3 presents a brief description of the data used (Section 3.1) and reports the results of the estimation (Section 3.2). The paper ends on the concluding note of Section 4.
